A practical guide to writing, and keeping alive, your own company rules for generative AI and AI agents. Covers basic policy and roles, the approved-AI list, information classes and how to confirm inference stays in-country, seven basic rules for AI agents, securing development and public services, devices, authentication and impersonation, technology crossing borders (export classification), incident response, and training and pledges. About 21 pages with model wording you can paste into a company rulebook, a 90-day drafting plan, FAQ, checklists and primary-source references. Written around Japanese law and guidance.